docs: fix memory layout comment in instructions sysvar

[dhl]

This patch fixes the inaccuracies in the memory layout description for the serialize_instructions function.

Specifically, it addresses incorrect byte offsets (e.g. incorrect starting offset for table of sets) and not taking variable number of accounts into consideration.

[Copilot]

Pull Request Overview

This PR fixes inaccuracies in the memory layout documentation for the serialize_instructions function in the instructions sysvar. The corrections address incorrect byte offsets and improve clarity by accounting for variable numbers of accounts in the serialization format.

Key changes:

• Corrected byte offset calculations throughout the memory layout comment
• Added explicit handling for variable account numbers in offset calculations
• Improved comment structure to show relative offsets within instruction and account sections

[joncinque]

Thanks for your contribution! This is definitely clearer, but I still had a slightly hard time with one part of it

[joncinque]

nit: I had a bit of a hard time reading these, do you think this might be clearer? We could also use square brackets, but that might make things worse

Suggested change

	// 2 + (33 * num_accounts)..34 + (33 * num_accounts) - program_id - 32 bytes
	// 34 + (33 * num_accounts)..36 + (33 * num_accounts) - data_len - 2 bytes
	// 36 + (33 * num_accounts)..36 + (33 * num_accounts) + data_len - data - variable length
	// (2 + (33 * num_accounts))..(34 + (33 * num_accounts)) - program_id - 32 bytes
	// (34 + (33 * num_accounts))..(36 + (33 * num_accounts)) - data_len - 2 bytes
	// (36 + (33 * num_accounts))..(36 + (33 * num_accounts) + data_len) - data - variable length

[dhl]

@joncinque thanks so much for reviewing this. I took your feedback and rewrote the instruction layout comment to improve clarity. I hope it is easier to follow now.

[joncinque]

Even better, thanks again!